Stop Searching. Start Trusting.

The curated directory of MCP servers, vetted for security, efficiency, and quality.

Browse DirectoryWhy Us?

Tired of the MCP "Marketplace" Chaos?

We built MCPScout.ai to solve the ecosystems biggest pain points.

No Insecure Dumps

We manually analyze every server for basic security flaws.

Easy Setup

Our gotcha notes warn you about complex setups.

Avoid "Token Hogs"

We estimate token costs for cost-effective agents.

Products, Not Demos

We filter out "Hello World" demos.

CATEGORIES:
SORT:

Vetted Servers(9120)

0
0
Medium Cost
Meshyarn icon

kairo

by Meshyarn

Sec7

A codebase intelligence and automation tool, providing functionalities for semantic search, code analysis, architectural understanding, and automated code modifications across multi-repository and multi-language environments.

Setup Requirements

  • ⚠️Requires downloading or pre-positioning large embedding models (e.g., 'multilingual-e5-small') from Hugging Face for local execution, which can consume significant disk space and bandwidth during initial setup.
  • ⚠️Relies on Tree-sitter WASM grammar files for AST parsing and a native Rust core module (`core_rs.node`) for performance-critical tasks. This may require specific build environments or pre-built binaries to be available for the target platform.
  • ⚠️Builds and maintains several persistent indexes (symbols, documents, embeddings, trigrams) on disk within a `.kairo` directory. For large codebases, this can consume substantial disk space and incur I/O overhead.
Verified SafeView Analysis
The server uses `child_process` for system-level commands (e.g., `prettier`, `grep`, custom alert commands in `AlertDispatcher`), which introduces a potential vector for command injection if inputs are not properly sanitized. Network activity is present for metrics/alerts but is not central to data processing. No direct use of `eval` or obvious hardcoded secrets. The architecture emphasizes offline-first operation, reducing external network attack surfaces for core logic.
Updated: 2026-01-19GitHub→
0
0
Low Cost
fairyto2 icon
Sec4

Provides a Model Context Protocol (MCP) service for AI assistants to securely manage SSH connections, execute commands, perform file operations, and maintain multi-turn interactive sessions on remote servers.

Setup Requirements

  • ⚠️Requires Python 3.13+
  • ⚠️Requires 'uv' package manager for installation and running commands.
  • ⚠️Default remote server config (config/remote_config.json) has authentication disabled; MUST be configured for secure use.
Review RequiredView Analysis
The default remote server configuration (`config/remote_config.json`) sets `enable_auth: false`, which is a critical security vulnerability for any network-exposed service. It also uses a placeholder `jwt_secret`. In `SSHConnectionManager`, `paramiko.AutoAddPolicy()` is used, automatically adding new host keys without verification, which is susceptible to Man-in-the-Middle attacks. Command execution directly uses input arguments from MCP requests, making robust authentication to the MCP server itself paramount. It's not safe for production use without configuring authentication and strong secrets.
Updated: 2026-01-19GitHub→
0
0
Medium Cost
mbrown007 icon

sm3_agent

by mbrown007

Sec6

An intelligent AI chat agent for multi-customer monitoring infrastructure, supporting natural language queries and proactive anomaly detection across Grafana, AlertManager, and Genesys Cloud.

Setup Requirements

  • ⚠️Requires OpenAI API Key (Paid)
  • ⚠️Docker & Docker Compose are essential for quick start and development setups
  • ⚠️Frontend requires backend server running on `http://localhost:8000`
Verified SafeView Analysis
The development Docker Compose configuration mounts `/var/run/docker.sock` into the agent container. This is a significant security risk as a compromise of the agent container could lead to host compromise. The agent also uses an LLM to interpret commands and interact with the monitoring infrastructure, which presents a general risk of prompt injection or unintended actions if the LLM generates malicious tool inputs. However, command execution is restricted by a configurable allowlist (`MCP_COMMAND_ALLOWLIST`) and defaults to a 'suggest' mode (`MCP_EXECUTION_MODE=suggest`), mitigating arbitrary command execution. Environment variables are validated using Pydantic, which is a good practice.
Updated: 2026-01-19GitHub→
0
0
High Cost
williamayal icon

wetrack-mcp-server

by williamayal

Sec2

Provides an MCP (Model Context Protocol) server for AI models to generate and execute MongoDB aggregation pipelines on financial event data.

Setup Requirements

  • ⚠️Requires an OpenAI API Key (paid service) for pipeline generation.
  • ⚠️Requires a running MongoDB instance and database URI.
  • ⚠️For remote access, requires Ngrok or similar tunnelling solution.
Review RequiredView Analysis
Critical vulnerability: The main '/mcp' endpoint, which handles AI tool calls, has its authentication mechanism explicitly commented out (disabled) in `src/server_http.py`, making it publicly accessible regardless of `.env` settings. OAuth2 and Bearer tokens are stored in a local JSON file (`oauth_tokens.json`) for persistence, which is insecure for production. CORS is set to allow all origins ('*'). The server executes LLM-generated MongoDB pipelines directly, posing a risk if the LLM is compromised or misaligned, as it could generate destructive operations (though aggregation views typically mitigate some direct write risks).
Updated: 2025-12-01GitHub→
0
0
Low Cost
Khushi-c-sharma icon
Sec9

Provides AI-powered budgeting, smart expense tracking, and affordability insights via the Model Context Protocol (MCP) for financial management.

Setup Requirements

  • ⚠️Requires Python 3.10 or higher.
  • ⚠️Requires specific configuration in Claude Desktop for integration, including an absolute path to the 'server.py' file.
Verified SafeView Analysis
The server uses parameterized queries for SQLite database interactions, effectively preventing SQL injection vulnerabilities. Input validation is present for categories, dates, and amounts across various tools. Sensitive data (budgets, expenses) is stored locally in SQLite and JSON files, which is secure for a single-user local application but requires secure system management. No hardcoded API keys or external network risks were identified as it runs via stdio transport (FastMCP). The flexible date/category parsing handles invalid inputs gracefully without crashing or exposing errors.
Updated: 2025-12-05GitHub→
0
0
Low Cost
Akungapaul icon

wp-advanced-mcp

by Akungapaul

Sec1

Manages WordPress installations by providing an MCP interface for WP-CLI commands, database operations, and WordPress REST API interactions.

Setup Requirements

  • ⚠️Requires a running WordPress installation.
  • ⚠️WP-CLI must be installed and accessible, with `WP_CLI_PATH` and `WORDPRESS_PATH` configured.
  • ⚠️Requires WordPress REST API credentials (username and app password) for REST client operations.
  • ⚠️SSH access details (`SSH_HOST`, `SSH_PORT`, `SSH_USER`, `SSH_KEY_PATH`) are required if WP-CLI is to be executed remotely via SSH.
Review RequiredView Analysis
The server exposes highly privileged operations like `execute_wp_cli` and `database_query` which allow execution of arbitrary WP-CLI commands and SQL queries (including destructive ones if `read_only` is not strictly enforced). There is no explicit authentication or authorization layer visible in the provided code, meaning any party with access to this MCP server endpoint could potentially execute these commands. This poses critical risks including system compromise, data loss, privilege escalation, and command/SQL injection if input sanitization by the underlying clients is not robust enough for all possible inputs.
Updated: 2025-11-28GitHub→
0
0
Low Cost

Custom-MCP-Server

by chrisgeorgeofficial

Sec7

Provides a custom implementation of a Minecraft (MCP) server.

Setup Requirements

  • ⚠️Python environment required.
  • ⚠️Dependencies must be installed via `pip install -r requirements.txt`.
  • ⚠️May require specific network configuration (e.g., port forwarding) for external access.
Verified SafeView Analysis
As a custom network server, it inherently poses risks if not properly secured or configured. Without access to the source code, specific vulnerabilities like 'eval' usage, obfuscation, or insecure network practices cannot be assessed. Assumes standard Python library usage and no malicious intent.
Updated: 2025-11-18GitHub→
0
0
Low Cost
ConnorH2582 icon
Sec8

A full-stack geospatial similarity search application allowing natural-language queries to find visually similar imagery chips on an interactive map with live weather.

Setup Requirements

  • ⚠️Requires `embeddings.db` file to be downloaded separately and placed in the project root, without which the backend will not run.
Verified SafeView Analysis
The server demonstrates good security practices for a demo application: parameterized queries are used for DuckDB to prevent SQL injection, CORS is correctly configured with `allow_credentials=False` when `allow_origins=["*"]`, and external API calls (OSM Nominatim, Open-Meteo) do not rely on hardcoded secrets (Open-Meteo is keyless, OSM is public). No 'eval' or obvious malicious code patterns were found. Minor areas for further hardening in a production environment could include explicit input validation for all API parameters, rate limiting, and robust authentication/authorization mechanisms.
Updated: 2025-12-05GitHub→
0
0
High Cost
roberto-delfiore icon

claude-code-best-practices

by roberto-delfiore

Sec9

This repository provides a comprehensive guide and configuration examples for leveraging Claude Code's advanced features, including Model Context Protocol (MCP) servers, sub-agents, hooks, and CI/CD integration, specifically tailored for professional use in enterprise environments.

Setup Requirements

  • ⚠️Requires an Anthropic account with a paid tier (Pro, Max, Teams, or Enterprise).
  • ⚠️Requires Node.js and npm/npx installed for setting up and running various MCP servers and hooks.
  • ⚠️Specific environment variables (e.g., API keys, database URLs) must be configured depending on the integrated services.
Verified SafeView Analysis
The provided source code is primarily documentation for Claude Code best practices, which heavily emphasizes a layered security model including user approval, sandboxing (filesystem and network isolation), and a robust permission system (allow/deny lists). It explicitly warns against hardcoded secrets, dangerous shell commands (e.g., `rm -rf`, `curl | bash`), and unauthorized file access, and suggests strict containerization for maximum security. There is no `eval`, obfuscation, or direct malicious code in the provided documentation snippets; rather, the documentation outlines how to configure for secure operation. The strong emphasis on security in the documentation is a significant positive.
Updated: 2026-01-17GitHub→
0
0
Medium Cost
marshally icon

mcp-refactoring

by marshally

Sec8

An MCP server that exposes Martin Fowler's refactoring catalog to LLMs through a pluggable, language-agnostic architecture for code transformation.

Setup Requirements

  • ⚠️Requires Python 3.10+.
  • ⚠️Requires 'molting-cli' (or another language-specific backend) to be installed and available in PATH for Python refactorings. This is an external dependency.
  • ⚠️Code analysis and structure inspection for Python are currently stubbed or return limited information, relying on future enhancements to the 'molting-cli' backend.
Verified SafeView Analysis
The server executes external language-specific CLI tools (e.g., 'molting-cli'). It uses safe subprocess execution (`asyncio.create_subprocess_exec` with `*args`) to prevent typical shell injection from the MCP server's side. The primary security consideration lies in the robustness of these external CLI tools to handle potentially malicious or malformed 'target' and 'params' inputs without introducing vulnerabilities themselves. No direct 'eval', 'exec', or hardcoded secrets were found, and it communicates via stdio, not network ports.
Updated: 2025-11-29GitHub→
0
0
Medium Cost
masseater icon

n8n-mcp-server

by masseater

Sec9

Enables AI models to manage and interact with n8n workflow automation platform through a standardized Model Context Protocol (MCP) interface.

Setup Requirements

  • ⚠️Requires an active n8n instance (local or remote) and a valid n8n API key for functionality.
  • ⚠️Requires Node.js 22.10.0 or higher.
  • ⚠️Tools like 'create_workflow_from_file' and 'replace_workflow_from_file' expect the AI agent to provide valid local file paths to workflow JSON definitions on the server where the MCP server is running.
Verified SafeView Analysis
The server correctly uses environment variables for sensitive API keys and implements basic retry logic for API calls. Logging masks API keys in output. File operations (creating/replacing workflows from local JSON files) assume the AI agent provides trusted paths; `path.resolve` is used but a malicious AI could potentially craft paths if not properly sandboxed. Overall, it is well-structured for its intended use case as a developer tool with a trusted AI agent.
Updated: 2025-12-04GitHub→
0
0
Medium Cost
Anishshah2-gmail icon

candidate-sourcing-mcp

by Anishshah2-gmail

Sec8

Provides an AI agent (like Claude Desktop) with tools for LinkedIn-based candidate sourcing, including searching, detailed profile fetching, bookmarking, and exporting.

Setup Requirements

  • ⚠️Requires LinkedIn Talent Solutions API access, which typically involves a corporate partnership/agreement and is not a simple API key purchase.
  • ⚠️Requires Node.js version 18 or higher to run.
  • ⚠️The Proxycurl data provider, while coded, has been shut down due to a LinkedIn lawsuit, making it non-viable for use.
Verified SafeView Analysis
The server uses `zod` for robust input validation. Sensitive API credentials are correctly handled via environment variables, reducing the risk of hardcoded secrets. Local bookmark storage is confined to the user's home directory. The primary security considerations stem from reliance on external APIs (LinkedIn Talent Solutions or deprecated Proxycurl) and the handling of potentially sensitive candidate data, which is an inherent risk of the application's purpose, not a flaw in the server's code itself.
Updated: 2025-12-03GitHub→
PreviousPage 334 of 760Next